LVM-Logical Volume Management

LVM-Logical Volume Management

===============================

LVM is a logical volume manager for the Linux kernel; it manages disk drives and similar mass-storage devices, in particular large ones. The term "volume" refers to a disk drive or partition.Every system contains Physical Volumes[PV]. Such as hard disks, partitions or external storage's. Volume management treats PVs as sequences of chunks called Physical Extents (PEs). There is an another concept also. Logical Extents(LE). Each LE maps one-to-one PE. The system pools LEs into a Volume Group (VG). We can extend this VG by adding a group of Logical extents to it from anywhere at any time.

Uses of LVM:

1.  Extending the partitions online

2.  Grouping of hard disks

3.  Reducing the partitions / hard disk size (offline)

4.  Increasing the performance

5.  Taking Backup (SNAPSHOT)

Example:

Here we are going to discuss a Volume Group(VG) created from 3 Physical Volumes(PV). And in that VG we'll create two Logical Volumes(LV) And mount it to /linux1 and /linux2 respectively.Now first of all in our example, we have 3 partitions. In real industry it may be 3 different hard disks.

Let it be (1) /dev/sda5 (2) /dev/sda6 (3) /dev/sda7 each of size 300Mb.

[root@vm4 ~]# fdisk -l

/dev/sda5 429 465 297171 8e Linux LVM

/dev/sda6 466 502 297171 8e Linux LVM

/dev/sda7 503 539 297171 8e Linux LVM

Steps:

First we will convert this partitions(hard disks) into Physical Volumes(PV). Then we'll create a Volume Group (VG)from those PV s. Then inside that VG, We'll create two Logical Volumes (LV) and we'll mount those for use.

Step1: Creating Physical Volume(PV)s

Partitions or disks can be converted into PV s using the following Command.

#pvcreate PARTITION_NAMES

#pvcreate /dev/sda5 /dev/sda6 /dev/sda7

or as below

#pvcreate /dev/sda{5,6,7}

[root@vm4 ~]# pvcreate /dev/sda5 /dev/sda6 /dev/sda7

Physical volume "/dev/sda5" successfully created

Physical volume "/dev/sda6" successfully created

Physical volume "/dev/sda7" successfully created

Monitoring or verifying the PV s:

You can verify the PV s using following commands,

#pvscan

#pvdisplay

#pvs

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 lvm2 -- 290.21M 290.21M

/dev/sda6 lvm2 -- 290.21M 290.21M

/dev/sda7 lvm2 -- 290.21M 290.21M

Step2: Creating Volume Group(VG):

The Physical Volumes are grouped into one to make it a Volume Group(VG). It can be done using the following command.

#vgcreate VG_NAME PV_NAMES

#vgcreate oracle /dev/sda5 /dev/sda6 /dev/sda7

or as below

#vgcreate oracle /dev/sda{5,6,7}

It will have a approximate size of 900(300+300+300). Some part will for writing headers LE and making LE-PE mapping.

[root@vm4 ~]# vgcreate oracle /dev/sda5 /dev/sda6 /dev/sda7

Volume group "oracle" successfully created

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

oracle 3 0 0 wz--n- 864.00M 864.00M

Monitoring or verifying the VG s:

You can verify the VG s using following commands,

#vgscan

#vgdisplay

#vgs

Output of #vgs is shown above the picture.

Step3: Creating Logical Volumes In Volume Group:

Now we got a volume group "Oracle" of size as the total size of all individual disks/partitions. Now we can create Logical Volumes or usable partitions inside it. We will create two logical Volumes lvm1 and lvm2 of size 100Mb each.

The lvm1 and lvm2 can be created using the following commands.

#lvcreate -L SIZE -n LV_NAME VG_NAME

#lvcreate -L 100M -n lvm1 oracle

#lvcreate -L 100M -n lvm1 oracle

-->

[root@vm4 ~]# lvcreate -L 100M -n lvm1 oracle

Logical volume "lvm1" created

[root@vm4 ~]# lvcreate -L 100M -n lvm2 oracle

Logical volume "lvm2" created

Monitoring or verifying the LV s:

You can verify the LV s using following commands,

#lvscan

#lvdisplay

#lvs

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 oracle -wi-a- 100.00M

lvm2 oracle -wi-a- 100.00M

The Logical Volumes lvm1 and lvm2 should be formatted(making file system in those) before mounting it. Then only you can use those partitions.

==>

Here formatting in ext3:

#mkfs.ext3 /dev/oracle/lvm1

#mkfs.ext3 /dev/oracle/lvm2

Making Mount Points:

#mkdir /linux1

#mkdir /linux2

Mounting (Temporary):

#mount /dev/oracle/lvm1 /linux1

#mount /dev/oracle/lvm2 /linux2

  

[root@vm4 ~]# mount

[Output truncated]

/dev/mapper/oracle-lvm1 on /linux1 type ext3 (rw)

/dev/mapper/oracle-lvm2 on /linux2 type ext3 (rw)

Extending a Logical Volume (Online):

Now We have a Volume group "oracle" of size about 900Mb. And two Logical vloumes lvm1 and lvm2 mounted on /linux1 and /linux2 respectively. Each having 100Mb size. Now we'll extend the size of lvm1 by 100Mb.

Extending size of a LV can be done online, That is by keeping them mounted. It can be achived by executing following command.

#lvextend -L +SIZE THE_PATH_OF_LV

#lvextend -L +100M /dev/oracle/lvm1

Before:

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 oracle -wi-ao 100.00M

lvm2 oracle -wi-ao 100.00M

Executing:

[root@vm4 ~]# lvextend -L +100M /dev/oracle/lvm1

Extending logical volume lvm1 to 200.00 MB

Logical volume lvm1 successfully resized

After:

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 oracle -wi-ao 200.00M

lvm2 oracle -wi-ao 100.00M

After executing above commands you can verify the changed size by any of following commands

#lvs, #lvdisplay, #lvscan

But if you check

#df -hT

it will be showing the old size only. Because the file system is updated the changed in Logical Volume. It can be updated by following command.

#resize2fs /dev/oracle/lvm1

NOTE: In case of extending the LV is resized first and the file system after that. But in case of shrinking a LV, file system is shrinked first followed by the shrink in LV.

Before:

[root@vm4 ~]# df -hT

Filesystem Type Size Used Avail Use% Mounted on

/dev/mapper/oracle-lvm1

ext3 97M 5.6M 87M 7% /linux1

/dev/mapper/oracle-lvm2

ext3 97M 5.6M 87M 7% /linux2

Executing:

[root@vm4 ~]# resize2fs /dev/oracle/lvm1

resize2fs 1.39 (29-May-2006)

Filesystem at /dev/oracle/lvm1 is mounted on /linux1; on-line resizing required

Performing an on-line resize of /dev/oracle/lvm1 to 204800 (1k) blocks.

The filesystem on /dev/oracle/lvm1 is now 204800 blocks long.

After:

[root@vm4 ~]# df -hT

Filesystem Type Size Used Avail Use% Mounted on

/dev/mapper/oracle-lvm1

ext3 194M 5.6M 179M 4% /linux1

/dev/mapper/oracle-lvm2

ext3 97M 5.6M 87M 7% /linux2

Shrinking a Logical Volume (Offline):

As we extended the size of Logical Volume, we can reduce the size also. But in later case, it can be done only offline. That is the LV should be unmounted for reducing its size.

For Shrinking a Volume:

1.  Filesystem must be reduced first

2.  Requires a filesystem check and cannot be performed online

3.  #lvreduce can then reduce the volume

Now We have a Volume group "oracle" of size about 900Mb. And two Logical volumes lvm1 and lvm2 mounted on /linux1 and /linux2 respectively. Size of lvm1 is 200Mb and size of lvm2 is 100Mb. Now we'll reduce the size of lvm2 by 40Mb.

Current status:

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 oracle -wi-ao 200.00M

lvm2 oracle -wi-ao 100.00M

Step1: Unmont the volume:

#umount /linux2

Step2: Checking the file system:

#e2fsck -f LV_Path

#fsck -f /dev/oracle/lvm2

[root@vm4 ~]# e2fsck -f /dev/oracle/lvm2

e2fsck 1.39 (29-May-2006)

Pass 1: Checking inodes, blocks, and sizes

Pass 2: Checking directory structure

Pass 3: Checking directory connectivity

Pass 4: Checking reference counts

Pass 5: Checking group summary information

/dev/oracle/lvm2: 11/25688 files (9.1% non-contiguous), 8914/102400 blocks

This is to make sure that the file system is in a consistent state.

Step3: Resizing the file system:

#resize2fs LV_Path Final_size

#resize2fs /dev/oracle/lvm2 60M

[Total size was 100Mb. Reduction 40Mb. So final size is 100-40=60]

[root@vm4 ~]# resize2fs /dev/oracle/lvm2 60M

resize2fs 1.39 (29-May-2006)

Resizing the file system on /dev/oracle/lvm2 to 61440 (1k) blocks.

The file system on /dev/oracle/lvm2 is now 61440 blocks long.

Step4: Now reduce the LV using #lvreduce

#lvreduce -L Size LV_Path

#lvreduce -L 60M /dev/oracle/lvm2

[root@vm4 ~]# lvreduce -L 60M /dev/oracle/lvm2

WARNING: Reducing active logical volume to 60.00 MB

THIS MAY DESTROY YOUR DATA (file system etc.)

Do you really want to reduce lvm2? [y/n]: y

Reducing logical volume lvm2 to 60.00 MB

Logical volume lvm2 successfully resized

Before Reducing:

root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 oracle -wi-ao 200.00M

lvm2 oracle -wi-ao 100.00M

After Reducing:

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 oracle -wi-ao 200.00M

lvm2 oracle -wi-a- 60.00M

After mounting:

[root@vm4 ~]# df -hT

/dev/mapper/oracle-lvm1 ext3 194M 5.6M 179M 4% /linux1

/dev/mapper/oracle-lvm2 ext3 59M 5.3M 50M 10% /linux2

Adding a Physical Volume to a VG:

Now we have a Volume Group "oracle" of size 900Mb. Suppose we used up to the maximum usable size. So we need to extend the size of the VG.

In this case we will create a new partition/a new hard disk, and will make it a Physical Volume and add it to the Volume group.

Current status:

 [root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

oracle 3 2 0 wz--n- 864.00M 624.00M

Now we created one more partition with id LVM.

/dev/sda5 429 465 297171 8e Linux LVM

/dev/sda6 466 502 297171 8e Linux LVM

/dev/sda7 503 539 297171 8e Linux LVM

/dev/sda8 540 576 297171 8e Linux LVM

Step1:

First we have to convert it to physical volume.

#pvcreate /dev/sda8

[root@vm4 ~]# pvcreate /dev/sda8

Physical volume "/dev/sda8" successfully created

Current status of all Physical Volumes:

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 oracle lvm2 a- 288.00M 88.00M

/dev/sda6 oracle lvm2 a- 288.00M 248.00M

/dev/sda7 oracle lvm2 a- 288.00M 288.00M

/dev/sda8 lvm2 -- 290.21M 290.21M

See /dev/sda8 is not the part of the VG oracle

Step2:

Now we will add the PV /dev/sda8 to Volume Group "oracle"

#vgextend -v oracle /dev/sda8

[ -v is for verbose. To see what is happening]

[root@vm4 ~]# vgextend -v oracle /dev/sda8

Checking for volume group "oracle"

Archiving volume group "oracle" metadata (seqno 5).

Wiping cache of LVM-capable devices

Adding physical volume '/dev/sda8' to volume group 'oracle'

Volume group "oracle" will be extended by 1 new physical volumes

Creating volume group backup "/etc/lvm/backup/oracle" (seqno 6).

Volume group "oracle" successfully extended

After extending the VG with new Physical Volume:

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr Vsize VFree

oracle 4 2 0 wz--n- 1.12G 912.00M

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 oracle lvm2 a- 288.00M 88.00M

/dev/sda6 oracle lvm2 a- 288.00M 248.00M

/dev/sda7 oracle lvm2 a- 288.00M 288.00M

/dev/sda8 oracle lvm2 a- 288.00M 288.00M

Now /dev/sda8 became a part of VG oracle.

Removing a Physical volume form a VG:

Before removing a physical volume from a volume group, you can make sure that the physical volume is not used by any logical volumes by using the #pvdisplay command. If the physical volume is still being used you will have to migrate the data to another physical volume using the #pvmove command. Then use the vgreduce command to remove the physical volume.

Current status of Pysical Volumes:

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 oracle lvm2 a- 288.00M 88.00M

/dev/sda6 oracle lvm2 a- 288.00M 248.00M

/dev/sda7 oracle lvm2 a- 288.00M 288.00M

/dev/sda8 oracle lvm2 a- 288.00M 288.00M

In this example we will remove the Pysical Volume /dev/sda5 in which some Physical Extents are already used by some LV. So They need to be migrated.

The status of /dev/sds5 is

[root@vm4 ~]# pvdisplay /dev/sda5

--- Physical volume ---

PV Name /dev/sda5

VG Name oracle

PV Size 290.21 MB / not usable 2.21 MB

Allocatable yes

PE Size (KByte) 4096

Total PE 72

Free PE 22

Allocated PE 50

PV UUID 9l5HlF-h8Of-2J6D-TDr4-BY34-cREh-7U5zxm

Step1:

Migrate the used Pes using #pvmove command

[root@vm4 ~]# pvmove -v /dev/sda5

Finding volume group "oracle"

Archiving volume group "oracle" metadata (seqno 6).

[output truncated]

Creating volume group backup "/etc/lvm/backup/oracle" (seqno 9).

Step2:

Now reduce the Volume Group size by command #vgreduce

#vgreduce VG_name Removing_PV_Path

#vgreduce oracle /dev/sda5

[root@vm4 ~]# vgreduce oracle /dev/sda5

Removed "/dev/sda5" from volume group "oracle"

After Remvoing PV:

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

oracle 3 2 0 wz--n- 864.00M 624.00M

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 lvm2 -- 290.21M 290.21M /Now not part of oracle VG

/dev/sda6 oracle lvm2 a- 288.00M 248.00M

/dev/sda7 oracle lvm2 a- 288.00M 88.00M

/dev/sda8 oracle lvm2 a- 288.00M 288.00M

Merging two Volume Groups:

Two different Volume Groups can be merged to a single Volume Group.

Suppose we have two VGs "oracle" and "linux".

Current status of VGs are:

VG #PV #LV #SN Attr VSize VFree

linux 1 0 0 wz--n- 288.00M 288.00M

oracle 3 2 0 wz—n- 864.00M 624.00M

Now we are going to merge "oracle" and "linux" to get a single VG linux.

We are using the command #vgmerge for this.

#vgmerge merges two existing volume groups. The inactive SourceVolumeGroupName will be merged into the DestinationVolumeGroupName if physical extent sizes are equal and physical and logical volume summaries of both volume groups fit into DestinationVolumeGroupName's limits.

#vgmerge -v databases my_vg

merges the inactive volume group named "my_vg" into the active or inactive volume group named "databases" giving verbose runtime information.

Step1:

we are going to merge "oracle" and "linux" to get a single VG linux.

Make the source Volume Group inactive. In this case source is oracle.

#vgchange -a n oracle

umount the LV which uses the oracle VG then only u can inactivate it.

[root@vm4 ~]# umount /linux1

[root@vm4 ~]# umount /linux2

[root@vm4 ~]# vgchange -a n oracle

0 logical volume(s) in volume group "oracle" now active

#vgmerge Destination Source

we are going to merge "oracle" and "linux" to get a single VG linux. So

#vgmaerge -v linux oracle

[root@vm4 ~]# vgmerge -v linux oracle

Checking for volume group "linux"

Checking for volume group "oracle"

Archiving volume group "oracle" metadata (seqno 10).

Archiving volume group "linux" metadata (seqno 1).

Writing out updated volume group

Creating volume group backup "/etc/lvm/backup/linux" (seqno 2).

Volume group "oracle" successfully merged into "linux"

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

linux 4 2 0 wz--n- 1.12G 912.00M

[root@vm4 ~]# vgchange -a y linux

2 logical volume(s) in volume group "linux" now active

[root@vm4 ~]# mount /dev/linux/lvm1 /linux1

[root@vm4 ~]# mount /dev/linux/lvm2 /linux2

[root@vm4 ~]# mount

[output truncated]

/dev/mapper/linux-lvm1 on /linux1 type ext3 (rw)

/dev/mapper/linux-lvm2 on /linux2 type ext3 (rw)

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 lvm2 -- 290.21M 290.21M

/dev/sda6 linux lvm2 a- 288.00M 248.00M

/dev/sda7 linux lvm2 a- 288.00M 88.00M

/dev/sda8 linux lvm2 a- 288.00M 288.00M

/dev/sda9 linux lvm2 a- 288.00M 288.00M

Splitting a Volume Group into two:

We can split a Volume Group into two Volume Groups using the command #vgsplit

We have one VG oracle with 4 PVs. We will split that VG oracle to oracle and another VG redhat. The PVs /dev/sda8and /dev/sda9 will be moved to redhat.

The syntax is as follows:

#vgsplit EXISTING_VG NEW_VG PATH OF PVs_TO _BE_MOVED

#vgsplit oracle redhat /dev/sda8 /dev/sda9

[root@vm4 ~]# vgsplit linux redhat /dev/sda8 /dev/sda9

New volume group "redhat" successfully split from "linux"

Before:

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

linux 4 2 0 wz--n- 1.12G 912.00M

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 lvm2 -- 290.21M 290.21M

/dev/sda6 linux lvm2 a- 288.00M 248.00M

/dev/sda7 linux lvm2 a- 288.00M 88.00M

/dev/sda8 linux lvm2 a- 288.00M 288.00M

/dev/sda9 linux lvm2 a- 288.00M 288.00M

After:

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

linux 2 2 0 wz--n- 576.00M 336.00M

redhat 2 0 0 wz--n- 576.00M 576.00M

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 lvm2 -- 290.21M 290.21M

/dev/sda6 linux lvm2 a- 288.00M 248.00M

/dev/sda7 linux lvm2 a- 288.00M 88.00M

/dev/sda8 redhat lvm2 a- 288.00M 288.00M

/dev/sda9 redhat lvm2 a- 288.00M 288.00M

[root@vm4 ~]# vgchange -a y linux

2 logical volume(s) in volume group "linux" now active

[root@vm4 ~]# vgchange -a y redhat

0 logical volume(s) in volume group "redhat" now active

Creating SNAPSHOT (Backup):

We can take the back up of any particular Logical Volumes. Snapshot will be stored in the Same volume group. Snapshot will take only 3-5% of the original size of Logical Volume. This is the current status of our machine:

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

lvm1 linux -wi-a- 200.00M

lvm2 linux -wi-a- 40.00M

[root@vm4 ~]# vgs

VG #PV #LV #SN Attr VSize VFree

linux 2 2 0 wz--n- 576.00M 336.00M

redhat 2 0 0 wz--n- 576.00M 576.00M

[root@vm4 ~]# pvs

PV VG Fmt Attr PSize PFree

/dev/sda5 lvm2 -- 290.21M 290.21M

/dev/sda6 linux lvm2 a- 288.00M 248.00M

/dev/sda7 linux lvm2 a- 288.00M 88.00M

/dev/sda8 redhat lvm2 a- 288.00M 288.00M

/dev/sda9 redhat lvm2 a- 288.00M 288.00M

We have two Logical Volumes lvm1 and lvm2 mounted on /linux1 and /linux2 respectively. We'll take the SNAPSHOT of the LV lvm1 which is mounted on /linux1 currently there are following files in /linux1 [i.e. in lvm1] 

[root@vm4 ~]# cd /linux1 

[root@vm4 linux1]# ls 

a b c lost+found 

Step1: Snapshot is actually a Logical Volume only. It will be saved in the same VG. We can create lvm snapshot using command #lvcreate with options for snapshot. The syntax is as follows 

#lvcreate --size SIZE --snapshot --name Name_Of_Snapshot Path_of_the_LV 

[root@vm4 linux1]# lvcreate --size 10M --snapshot --name snap /dev/linux/lvm1 

Rounding up size to full physical Logical volume "snap" created 

[root@vm4 linux1]# lvs LV VG Attr LSize Origin Snap% Move Log Copy% Convert lvm1 linux owi-ao 200.00M lvm2 linux -wi-ao 40.00M snap linux swi-a- 12.00M lvm1 0.10

[root@vm4 linux1]# vgs VG #PV #LV #SN Attr VSize VFree 

linux 2 3 1 wz--n- 576.00M 324.00M redhat 2 0 0 wz--n- 576.00M 576.00M

Now let’s mount the Snapshot and check whether it has everything from lvm1. Mounting snap to /snapshot

[root@vm4 linux1]# mkdir /snapshot 

[root@vm4 linux1]# mount /dev/linux/snap /snapshot 

[root@vm4 linux1]# cd /snapshot 

[root@vm4 snapshot]# ls 

a b c lost+found 

It has the same and all the contents of lvm1 

How to create mirrored lvm volumes: 

We can create mirrored lvm using the command #lvcreate with -m option. The syntax is as follows: 

#lvcreate -L Size -m1 -n Name_LV Name_VG Mirror_1st_leg   Mirror_2nd_leg   Log_Device -m1 means it’s a mirrored one type. So it will have one original[1st leg] one mirror[2nd leg] and a logging device for sync. 

#lvcreate -L 100M -m1 -n mlvm redhat  /dev/sda5  /dev/sda6  /dev/sda7 

[root@vm4 ~]# lvcreate -L 100M -m1 -n mlvm redhat /dev/sda5 /dev/sda6 /dev/sda7 

/dev/cdrom: open failed: Read-only file system Logical volume "mlvm" created 

[root@vm4 ~]# lvs LV VG Attr LSize Origin Snap% Move Log Copy% Convert 

mlvm redhat mwi-a- 100.00M mlvm_mlog 100.00 

How to create stripped lvm volumes: 

We can create lvms in stripped manner so that it will increase the performance. It can be done using the commad #lvcreate with -i option.

The syntax is as follows:

#lvcreate -L Size -i2 -n Name_LV Name_VG 1st _LV 2nd _LV

[root@vm4 ~]# lvcreate -L 100M -i2 -n slvm redhat /dev/sda8 /dev/sda9

Using default stripesize 64.00 KB

/dev/cdrom: open failed: Read-only file system

Rounding size (25 extents) up to stripe boundary size (26 extents)

Logical volume "slvm" created

[root@vm4 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy% Convert

slvm redhat -wi-a- 104.00M

For extending a stripped lvm you need to extend the VG with two different PVs . Else it won’t work.